Health condition prediction method of the computer numerical control machine tool parts by ensembling digital twins and improved LSTM networks

Abstract The advent of digital twin technology has precipitated a surge of interest in its implementation within the manufacturing sector, particularly with respect to enhancing productivity and mitigating the risk of equipment failure. However, due to the incomplete maturity of the technology, the manufacturing industry is prone to failures and reduced production efficiency in the production process. To reduce the frequency of failure and improve production efficiency, the optimized digital twin technology is combined with the long and short-term memory network to predict the health condition of the improved computer numerical control (CNC) machine tool. The proposed model is predicated on a convolutional neural network and a long and short-term memory network, the purpose of which is to extract the feature data of CNC machine tool parts. Second, the attention mechanism is introduced to enhance the importance weight of the model to the time feature. Finally, the Kalman filtering is employed to denoise the data and generate the final prediction result. The experimental findings of the model demonstrated that the maximum error between the predicted and true value of the convolution-long short-term memory (LSTM) model in the machine tool part health monitoring data was 0.06, and the minimum error was 0.035. The convolution-LSTM model in the anomalous point inspection displacement of the amount of positive displacement did not exceed 25 mm, and the negative displacement did not exceed −23 mm. The Kalman wave-LSTM model demonstrated an accuracy of 90.5% in monitoring the health status of the part at the minimum. The results show that the proposed method has good fitting effect and high accuracy in different monitoring cycles. This research provides a new method for monitoring and predicting the health condition of parts of CNC machine tools, which has important practical significance for optimizing and improving the production efficiency of manufacturing industries.